Skip to content

README_EN.md

Latest commit

 

History

History
191 lines (129 loc) · 5.16 KB

README_EN.md

File metadata and controls

191 lines (129 loc) · 5.16 KB
# Wise-will Edu Agent Platform

## 📚 Open-Source Educational AI Agent

This is an **open-source educational AI Agent product**.  
Its design is **inspired by excellent products such as Ant Lingguang, Doubao Aixue, and Feixiang Teacher**,  
and it is dedicated to **visually explaining everything you ask**, truly achieving the goal of being **built for education**.

---

### 🎯 Why Education?

We believe that **education is one of the most important real-world application scenarios for AI agents**.  
By choosing **open source**, we hope to:

- 🌱 Support more **schools**
- 👩‍🎓 Empower more **students**
- 👨‍🏫 Assist more **educators**

So that **AI-powered education** is not limited to a few platforms, but can be genuinely adopted and used by more people.

---

### 🚧 Project Status

- The current version is still **relatively rough**
- Features and user experience are **actively being refined**
- The project will **continue to evolve and improve**

---

### 🤝 Collaboration & Contribution

You are welcome to:

- Use it  
- Share feedback  
- Contribute code  

Let’s build better educational AI together 👏

---

> **Enjoy education agent time together 🚀**

---

## Features

- **Conversational Learning** – Enter a topic you want to learn, and the AI responds with guided, instructional dialogue
- **Concept Map Navigation** – Visualized knowledge structures with clickable nodes to explore related concepts
- **Real-time Visualization Generation** – The AI automatically generates interactive HTML visualizations
- **Cloud Deployment** – One-click deployment to the cloud via EdgeOne Pages MCP
- **Progress Tracking** – Real-time display of AI processing stages (Thinking → Writing → Tool Calling → Deployment)

---

## Examples

(See screenshots above)

---

## Quick Start

### Requirements

- Node.js >= 18
- Python >= 3.13
- pnpm

### Installation

```bash
# Clone the repository
git clone --recursive https://github.com/your-username/edu-ai-platform.git
cd edu-ai-platform

# Frontend dependencies
cd frontend && pnpm install

# Backend dependencies
cd ../backend && uv sync

Configuration

Copy the environment variable template and fill in your API key:

cp backend/.env.example backend/.env

Edit backend/.env:

# Use Kimi (default)
KIMI_API_KEY=sk-your-api-key
KIMI_MODEL_NAME=kimi-k2-turbo-preview

# Or use DeepSeek / OpenAI-compatible API
LLM_PROVIDER_TYPE=openai_legacy
OPENAI_API_KEY=sk-your-api-key
OPENAI_BASE_URL=https://api.deepseek.com
OPENAI_MODEL_NAME=deepseek-chat
OPENAI_REASONING_KEY=reasoning_content

Run

# Terminal 1 - Backend
cd backend && python main.py

# Terminal 2 - Frontend
cd frontend && pnpm dev

Visit: http://localhost:3000

Project Structure

edu-ai-platform/
├── frontend/           # Next.js frontend
│   └── src/
│       ├── app/        # Pages
│       ├── components/ # UI components
│       └── hooks/      # WebSocket & state management
├── backend/            # FastAPI backend
│   ├── main.py         # Main application
│   ├── kimi_runner.py  # Kimi CLI wrapper
│   ├── mcp.json        # MCP configuration
│   └── agent/          # Agent configs and prompts
├── kimi-cli/           # AI Agent core (submodule)
└── kosong/             # LLM abstraction layer (submodule)

Tech Stack

Layer Technology
Frontend Next.js 16, React 19, TypeScript, Tailwind CSS 4
Backend FastAPI, Python 3.13, WebSocket
AI Kimi CLI, MCP (Model Context Protocol)
Deployment EdgeOne Pages

API

Method Endpoint Description
GET / Service status
GET /health Health check
GET /api/history Fetch history
DELETE /api/history/{task_id} Delete history
WebSocket /ws/chat Real-time chat

Acknowledgements

This project is built upon the following excellent open-source projects:

  • Kimi CLI – An open-source coding agent framework by Moonshot AI, providing core capabilities such as tool invocation and MCP integration
  • Kosong – An open-source LLM abstraction layer by Moonshot AI that unifies message structures and supports multiple providers, making agent development simpler and more flexible
  • EdgeOne Pages MCP – An MCP service for EdgeOne Pages that enables one-click cloud deployment

Special thanks to the Moonshot AI team and EdgeOne Pages!

Contact

For questions or suggestions, please contact: 3038880699@qq.com

License

Apache-2.0 https://www.apache.org/licenses/LICENSE-2.0